load_only, dump_only for Nested fields

Question

Is there any way in marshmallow-sqlalchemy to specify load_only or dump_only fields for Nested (foos) when serializing/deserializng Bar?

class FooSchema(BaseSchema):
    class Meta(BaseSchema.Meta):
        model = Foo
        fields = ('id', 'name', 'date', 'clients')


class BarSchema(BaseSchema):
    class Meta(BaseSchema.Meta):
        model = Bar
        fields('id',)
    foos = Nested(FooSchema, many=True, only=('id', 'name'))
# is there a way to add to foos field something like load_only=('id',)
# without changing FooSchema?

score 3 · Accepted Answer · answered May 14 '19 at 13:44

I recommend against specifying only in the definition of a Nested relationship. Use exclude to prevent circular references and specify which fields you want only explicitly each time you are serializing.

Additionally, you generally shouldn't need to specify fields - marshmallow-sqlalchemy gives you that for free for most fields. Here's how I'd refactor the above:

class FooSchema(BaseSchema):
    bar = Nested('myproject.schemas.bar.BarSchema', exclude=('foos',))
    class Meta(BaseSchema.Meta):
        model = Foo
        dump_only = ('id',)


class BarSchema(BaseSchema):
    class Meta(BaseSchema.Meta):
        model = Bar
        dump_only = ('id',)
    # don't specify `only` here:
    foos = Nested(FooSchema, many=True, exclude=('bar',))

But how to add `only` for foos nested field during Foo serialization? — maslak, May 14 '19 at 15:32
You use `only` at the implementation-level when initializing your schema. For example: `barschema = BarSchema(only=['id', 'foos.id',])` — Stephen Fuhry, May 15 '19 at 16:08

load_only, dump_only for Nested fields

1 Answers1